Concept Drift Adaptation Using Self-Supervised and Reinforcement Learning In Android Malware Detection
arXiv:2605.24294v1 Announce Type: cross Abstract: Android malware detectors often degrade after deployment because of concept drift, while full retraining at each maintenance step is costly. We propose a chronological adaptive maintenance framework that models deployment-time maintenance as a sequential decision problem. The framework learns a stable latent representation through self-supervised learning during initialization, freezes the encoder, measures latent drift in the fixed representation space, and performs lightweight downstream adaptation using a trainable adapter and classification
The proliferation of Android devices and the sophistication of malware necessitate adaptive and cost-effective detection methods, pushing research into AI-driven solutions for real-world deployment challenges.
Organizations relying on AI for cybersecurity face significant operational costs and performance degradation due to concept drift; this research offers a framework to maintain system efficacy with reduced overhead.
The proposed framework enables more robust and economically feasible long-term deployment of AI-powered Android malware detection systems by automating adaptation and reducing manual intervention.
- · Cybersecurity providers
- · Android users
- · Organizations using AI for security
- · Reinforcement learning researchers
- · Malware developers
- · Traditional signature-based antivirus solutions
Reduced operational costs for maintaining AI security systems and improved detection rates against evolving threats.
Increased trust in AI-driven cybersecurity solutions, potentially leading to broader adoption across other threat vectors.
A shift in cyber warfare where defenders can more effectively adapt to new attack methodologies, creating a higher barrier for sophisticated adversaries.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG